OBJECTIVES:

With the increasing recognition of arthropathy as a major cause of morbidity in haemophilia comes a heightened interest in including long-term joint health outcomes in clinical assessment. The objective was to cross-validate a de-novo disease-transition model, to estimate the long-term accuracy of clinical outcome projections for treatment interventions in people with haemophilia.

METHODS:

A Markov disease-transition model focusing on the progression of haemophilic arthropathy, as estimated by the radiological Pettersson Score (PS), was developed. Health states characterize death and tunnel-like states of PS. Progression through PS is triggered by events of joint bleeds (JB), according to literature estimates from Coppola and colleagues [Coppola et al., 2017] (1 point increase in PS per 13.52 and 6.52 JB in patients under and over 25 years, respectively). Our model estimations of PS and cumulative JB were compared with those obtained by Fischer et al. [Fischer et al., 2011] via the coefficient of determination (R²) in both the prophylactic and on-demand arms. Baseline characteristics, clinical parameters, including annualized bleeding rates, and simulated treatment strategies were set to be the same as those depicted by Fischer and colleagues, which were based on data from a longitudinal cohort of severe haemophilia A patients (N = 180) without inhibitors to factor VIII.

RESULTS:

Our model explained by 97.4% and 98.0% the prophylaxis and episodic arms PS reported by Fischer and colleagues, respectively. Similarly, cumulative evolutions of JB reported by Fischer et al. were close to our own model predictions, with R² of 95.5% in both prophylaxis- and episodic-treated patients.

CONCLUSIONS:

Overall, our model provided joint health estimates consistent with alternative long-term projections previously outlined in the literature. Thus, the model can provide valuable information for the long-term prediction of arthropathy severity in people with haemophilia for different treatment strategies.